This invention concerns devices for sensing the relative intensity of electromagnetic radiation, such as light, at a number of locations and conversion of that information into electronic form for use in various apparatus. Examples of apparatus that can usefully incorporate such devices include photocopiers, facsimile transmitters, optical character readers and pattern recognition apparatus.
It is well known that an image on a document or other medium can be converted into electrical signals by illuminating the medium and measuring the amount of light reflected from, or transmitted through, each of a number of points on the image. Typically, a linear photoresponsive sensor is disposed a small distance from a document or image. The image is moved relative to the sensor as the sensor scans the image transversely to the direction relative motion, one line at a time. The photoresponse to the illumination is an electrical signal containing information indicative of the amount of light detected at numerous points along the scanned line. The electrical signal can be further processed to store or reproduce the image scanned in a modified or unmodified form, to transmit the information gathered from the image, to determine whether the information meets a test (pattern recognition) or denotes a known configuration (optical character reading).
Typical linear sensors operating in large format, i.e. over a large area, employ charge coupled devices or an array of photosensors. In large format applications, charge coupled devices require lenses, adding to the cost of the apparatus and introducing loss of optical resolution. In known arrays, each photosensor, usually a diode, is interrogated while or after it senses light intensity. Typically, the light intensity is determined by sensing changes in the electrical charge on each photosensor or on a capacitor connected in parallel with the photosensor. In general, the presence of light causes current flow and reduces the amount of stored charge. The amount of the stored charge that is lost is determined by measuring the quantity of charge required to restore the original charge. A separate component, such as a field effect transistor, is used to address each photosensor in each cycle or scan of the sensor. The field effect transistors themselves are typically connected to a shift register that collects the sensor data for further processing. These circuits are relatively complex since a separate switch is associated with each photoresponsive device and a shift register controls those switches. The switches, typically transistors, add a noise signal when they open and close that can be very troublesome since the signals sought to be detected are very weak. Moreover, the transistors have gate lengths of only a few microns. It is difficult to control accurately the positioning of a large number of these gates over a large area photosensor array using ordinary photolithography techniques.
The known photoresponsive arrays require a large number of connections to their shift registers. Many conductors must cross each other without being in electrical contact. This requirement means that either hybrid, i.e. not monolithic, circuits must be used to build the photosensor arrays, or a great number of conductor crossovers must be fabricated in a monolithic version. In a hybrid circuit array, electrical connections can be made by bonding wires to contact points so that the wiring is three dimensional, avoiding conductor crossovers. But assembly of hybrid circuits is not readily automated and is therefore expensive. The use of a large number of conductor crossovers in a monolithic circuit increases costs because extra processing steps are required and large numbers of crossovers adversely affect yield and reliability.
Examples of photosensitive arrays are disclosed in Dym and Hall U.S. Pats. Nos. 3,432,670 and 3,448,275. In the patent to Dym, pairs of diodes, one photoresponsive, one not photoresponsive, are connected in series. A number of the pairs are connected in parallel, separated by series connected resistors. The state of the photodiodes is determined by scanning the series-parallel array with a ramp voltage. As the ramp amplitude increases, each non-photoresponsive diode is forward biased in sequence. Each diode then remains forward biased until the end of the ramp is reached. The photoresponsive diodes are reverse biased by the ramp voltage. If any of the photodiodes are not illuminated only leakage current flows through them. As a result, they act like capacitors and store charge. The stored charge is retained until the photodiode is illuminated since no path for discharge is presented. Therefore, the residual voltages from the first voltage scan may distort all subsequent scans. This distortion might not be detrimental if the scanner senses only black and white images; but if a scanner is to sense a scale of greys intermediate to totally black and totally while states, then the residual charge will prevent accurate, reliable performance by the Dym scanner.
The Hall scanner is similar to that of Dym except that no monolithic embodiment is disclosed. Hall adds a capacitor across each photodiode which aids in detecting when the zero voltage point across the non-photoresponsive is reached. However, like the Dym scanner, the Hall scanner is susceptible to residual charge effects once the first voltage ramp has been applied since no current discharge path for unilluminated diodes is included.
Accordingly, what is needed is a photoresponsive array, preferably of monolithic construction, that is capable of coverting large area images into electrical signals, that does not require lenses, switches or a significant number of conductor crossovers and in which a scan of the sensors does not produce residual effects that adversely affect subsequent scans, so that black, white, and a multiple step gray scale image intensities can be detected.